Presentation Notes

The base scenario in this report simulates a network of 200 farms, with around 50 chickens per farm.
Each simulation creates a randomly generated network of farms.
Each simulation seeds one initial infected chicken at a random farm.
No culling practices are implemented.
The two GIFs below are examples of the base network scenario:
This graphic shows a simulation run that resulted in a long epidemic (80 days).
Nodes represent farms, with edges illustrating connections between farms.
Black nodes represent farms with a normal chicken population, while red nodes indicate farms with at least one infected chicken.

This graphic, on the other hand, shows a short duration simulation, where a seeded infection fails to spread throughout the network.

Scenario	Mean Proportion of Chickens Lost	Mean Proportion of Farms Infected	Mean Duration of Epidemic (days)	Mean Exposure Index	Proportion of Simulations that Failed to Spread
Base Scenario	0.0964127	0.0998200	36.08100	0.0004667	0.28
Base Scenario - Spread Only	0.1335270	0.1366944	48.55556	0.0006463	0.00

The following graphs show that there does not appear to be a substantial relationship between increases in the number of farms in the network and our epidemic summary statistics.
However, as the number of chickens increase in each farm, there are noticeable patterns of change for epidemic duration, proportion of infected farms, proportion of chickens lost across the network, and fraction of possible exposure measure.
Due to the scalability of this model across network size, subsequent simulations are run with a network size of 200 farms of 50 chickens.
The duration of the epidemic is affected by the number of chickens in each farm, but not the number of farms in the network.

The proportion of infected farms in a network is affected by the number of chickens in each farm, but not the number of farms in the network.

The proportion of chickens lost to infection or culling in a network is affected by the number of chickens in each farm, but not the number of farms in the network.

The fraction of infectious exposure is affected by the number of chickens in each farm, but not the number of farms in the network.

In the random growth scenario, 11% of farms grow from around 50 chickens to 500 chickens, resulting in a doubling of the total network chicken population. The location of these farms in the network is chosen at random.
Each simulation is seeded by choosing a random chicken to be infected. Since larger farms represent about 50% of the chicken population, there is about a 50% chance that the seeded infection will be on a larger farm.
The network graph below illustrates this growth scenario, with a random selection of nodes growing by a factor of 10 (indicated in green).

Scenario	Mean Proportion of Chickens Lost	Mean Proportion of Farms Infected	Mean Duration of Epidemic (days)	Mean Exposure Index	Proportion of Simulations that Failed to Spread
Growth in Random Farms	0.5917104	0.5740800	62.41400	0.0028588	0.202
Growth in Random Farms - Spread Only	0.7414420	0.7181328	77.50376	0.0035821	0.000

In the localized growth scenario, 11% of farms grow from around 50 chickens to 500 chickens, resulting in a doubling of the total network chicken population. The farms chosen to grow are adjacent nodes in a network, simulating the intensification of poultry production in one geographic, closely connected area.
Each simulation is seeded by choosing a random chicken to be infected. Since larger farms represent about 50% of the chicken population, there is about a 50% chance that the seeded infection will be on a larger farm.
The network graph below illustrates this growth scenario, with a clustered selection of nodes growing by a factor of 10 (indicated in green).

Scenario	Mean Proportion of Chickens Lost	Mean Proportion of Farms Infected	Mean Duration of Epidemic (days)	Mean Exposure Index	Proportion of Simulations that Failed to Spread
Base Scenario	0.0964127	0.0998200	36.08100	0.0004667	0.280
Base Scenario - Spread Only	0.1335270	0.1366944	48.55556	0.0006463	0.000
Growth in Random Farms	0.5917104	0.5740800	62.41400	0.0028588	0.202
Growth in Random Farms - Spread Only	0.7414420	0.7181328	77.50376	0.0035821	0.000
Localized Farm Growth	0.5127656	0.3995100	49.99800	0.0024792	0.202
Localized Farm Growth - Spread Only	0.6424679	0.4993734	61.76817	0.0031063	0.000

Culling is implemented as a two-step process:
- Farms are first identified to be culled if a critical number of chickens have died in a certain time window.
- Once a farm is identified, it is culled with an average speed parameterized by the culling time variable. When a farm is culled, all chickens (Susceptible, Infectious, and Recovered) are removed from that farm.
The following graphs show how variation in the cull time parameter effects chicken loss, epidemic spread, duration of epidemic, and infected chicken exposure.

Culling Time (days)	Mean Proportion of Chickens Lost	Mean Proportion of Farms Infected	Mean Duration of Epidemic (days)	Mean Exposure Index
1	0.2908635	0.177020	33.435	0.0008672
2	0.3992773	0.279585	41.311	0.0014814
3	0.4707156	0.355515	47.513	0.0020148
4	0.4932861	0.393805	50.212	0.0023617
5	0.5103106	0.421755	51.178	0.0026519
6	0.5585069	0.466455	54.654	0.0030650
7	0.5545843	0.471915	54.165	0.0032188
8	0.5915634	0.515060	58.297	0.0035533
9	0.5874925	0.514655	58.191	0.0036713
10	0.6115625	0.539915	60.246	0.0039618
11	0.6172856	0.550085	60.986	0.0040879
12	0.6222489	0.555785	60.918	0.0042105
13	0.6376711	0.576205	63.252	0.0044178
14	0.6345393	0.573060	63.312	0.0044858
15	0.6567254	0.596925	65.635	0.0047094
16	0.6323005	0.579120	63.062	0.0045832
17	0.6619602	0.608130	66.779	0.0048735
18	0.6593961	0.604180	68.279	0.0048979
19	0.6664469	0.614625	66.938	0.0050035
20	0.6500229	0.601185	66.890	0.0049526
21	0.6498694	0.602360	66.280	0.0049984